To build AI technology, startups are turning to Bigger Rivals for help

The tech industry loves garage startup stories. From Hewlett-Packard to Google, tales of innovative companies turned giants have inspired generations of entrepreneurs.

But the massive amounts of money and computing power needed for startups trying to make it happen with today’s cutting edge technology, and the artificial intelligence used in chatbots like ChatGPT and Google Bard, may make these inspiring tales a thing of the past.

In 2019, Aidan Gomez and Nick Frosst left Google to create an AI startup in Toronto called Cohere that could compete with their previous employer. Several months later, they went back to Google and asked if it would sell them the massive computing power they would need to build their AI technology. After Google CEO Sundar Pichai personally approved the arrangement, the tech giant gave them what they wanted.

“It’s the ‘Game of Thrones.’” David Katz, partner at Radical Ventures, Cohere’s first investor, said, “That’s what it is.” He added that big companies like Google, Microsoft and Amazon control the chips. “They choose who gets it.”

Creating a leading AI company is difficult without the support of the “Formidables”, who control huge data centers capable of running AI systems. And that has put industry giants in the driving seat – once again – of what many expect to be the most significant shift in the tech industry in decades.

OpenAI, the startup behind ChatGPT, recently raised $10 billion from Microsoft. Most of that money will be pumped back into Microsoft as it pays for time on the huge clusters of computer servers that the larger company runs. These machines, spanning thousands of specialized computer chips, are essential to improving and extending the skills of ChatGPT and similar technologies.

Competitors can’t keep up with OpenAI unless they get similar amounts of computing power. Cohere recently raised $270 million, bringing its total funding to more than $440 million. Much of that money will be used to buy computing power from the likes of Google.

Other startups have made similar arrangements, most notably a Silicon Valley company called Anthropic, which was founded in 2021 by a group of former OpenAI researchers. Character.AI, which was founded by two senior researchers from Google; and Inflection AI, which was founded by a former Google executive. Inflection raised a $1.3 billion funding round last week, bringing its total to $1.5 billion.

At Google, Mr. Gomez was part of a small research team that designed Transformerswhich is the underlying technology used to create chatbots such as ChatGPT and Google Bard.

Transformer is a powerful example of what scientists call a neural network – a mathematical system that can learn skills by analyzing data. Neural networks have been around for years, helping to drive everything from talking to digital assistants like Siri to instant translation services like Google Translate.

The converter took the idea into new territory. By running hundreds or even thousands of computer chips, it can analyze more data more quickly.

Using this technology, companies like Google and OpenAI have begun building systems that learn from vast amounts of digital text, including Wikipedia articles, digital books, and chat logs. As these systems analyze more and more data, they have learned to generate text on their own, including term papers, blog posts, poetry, and computer code.

These systems – called large language paradigms – now support chatbots like Google Bard and ChatGPT.

Long before ChatGPT arrived, Mr. Gomez left Google to start his own company alongside Mr. Frosst and another Toronto entrepreneur, Ivan Zhang. The goal was to build large language models that would rival Google’s.

At Google, he and his colleagues had access to nearly unlimited amounts of computing power. After leaving the company, he needed something similar. So he and his fellow co-founders bought it from Google, which sells access to the same chips through its cloud computing services.

Over the next three years, Queer built a large language model Almost no other competitors. Now, it sells this technology to other companies. The idea is to provide any company with the technology they need to build and run their own AI applications, from chatbots to search engines to personal tutors.

“The strategy is to build a platform that others can build on and experience,” Mr. Gomez said.

OpenAI offers a service along similar lines called GPT-4, which many companies are already using to build chatbots and other applications. This new technology can parse, create, and edit text. But it will soon be dealing with images and sounds, too. OpenAI is making a GPT-4 version that can instantly scan and describe an image and even answer questions about it.

Microsoft CEO Satya Nadella said the company’s arrangement with OpenAI is the kind of mutually beneficial relationship it has long nurtured with smaller competitors. “I grew up in a company that always made these kinds of deals with other companies,” he told the New York Times earlier this year.

As the industry races to match GPT-4, entrepreneurs, investors, and pundits debate who will ultimately be the winner. Most agree that OpenAI is the leader in this field. But Queer and a small handful of other companies are building similar technologies.

The tech giants are in a strong position because they have the massive resources needed to push these systems further than anyone else. Google too Patented transformerthe foundational technology behind the AI ​​systems that Cohere and many other companies are building.

But there is a wild card: open source software.

Meta, another behemoth with the computing power needed to build the next wave of AI, has just opened its latest big language model, which means anyone can reuse it and build on top of it. Many in the field believe that this kind of freely available software will allow anyone to compete.

“Having the collective minds of every researcher on earth would defeat any company,” said Amr Awadallah, CEO of AI startup Vectara and former CEO of Google. But they will still need to pay for access to a much larger competitor’s data centers.